• Chaos Theory
  • Posts
  • 🥟 Chao-Down #150 Apple already testing internal GPT chatbot, OpenAI tackles content moderation with GPT4, Meta's AI agents learn by copying toddlers

🥟 Chao-Down #150 Apple already testing internal GPT chatbot, OpenAI tackles content moderation with GPT4, Meta's AI agents learn by copying toddlers

Plus, China's Cyberspace Administration releases "interim" rules for regulating use of AI.

OpenAI just released a new blog exploring the use of AI for content moderation. In particular, they state:

We use GPT-4 for content policy development and content moderation decisions, enabling more consistent labeling, a faster feedback loop for policy refinement, and less involvement from human moderators.

Normally a task for human reviewers, content moderation has been a thorny problem for the largest social platforms. We’ve seen many stories of the grueling working conditions that companies have set up, just so that our timelines are free from inappropriate content.

But can AI really help solve what often is an area filled with grey areas and moral judgments?

We’ll have to see. 

-Alex, your resident Chaos Coordinator.

What happened in AI? 📰

Apple’s GPT chatbot is already in use internally (The Verge)

Google’s AI search experience adds AI-powered summaries, definitions and coding improvements (TechCrunch)

Can ChatGPT become a content moderator? (Semafor)

Meta’s AI Agents Learn to Move by Copying Toddlers (IEEE Spectrum)

China's Cyberspace Administration Releases "Interim" Rules Regulating the Use of Generative AI (JDSupra)

Google's next big swing at AI rumored to launch this fall (Android Police)

Workers are putting AI skills on resumes as demand heats up (Axios)

Always be Learnin’ 📕 📖

TheAlgorithms/Python: All Algorithms implemented in Python (Github)

Ask like a human: Implementing semantic search on Stack Overflow (Stack Overflow Blog)

AI and Memory Wall. “Several AI applications are becoming bottlenecked by intra/inter-chip and communication across/to AI accelerators” (Medium)

Projects to Keep an Eye On 🛠

Platypus - Quick, Cheap, and Powerful Refinement of LLMs (platypus-llm.github.io)

WizardLM/WizardMath Empowering Mathematical Reasoning for Large Language Models via Reinforced Evol-Instruct (Github)

ThousandBirdsInc/chidori: A reactive runtime for building durable AI agents (Github)

The Latest in AI Research 💡

V-DETR: DETR with Vertex Relative Position Encoding for 3D Object Detection (arxiv)

FactoDeepLearning/MultitaskVLFM: Leveraging Vision-Language Foundation Models for Fine-Grained Downstream Tasks (Github)

Exploration-Lab/IL-PCR: U-CREAT: Unsupervised Case Retrieval using Events extrAcTion (Github)

/Images/ucreat_pipeline.png

The World Outside of AI 🌎

Amazon Trims Packaging, Offers 'Ships in Own Container' Option (Entrepreneur)

Netflix finally streams video games, too - The Verge

China new app regulations to hit foreign developers (qz.com)

Mexico’s army-run airline due to start operations this year (qz.com)

How can AI help with elder care in the home? (qz.com)

The Gamification of Reading Is Changing How We Approach Books (shondaland.com)

Real estate markets scramble following cyberattack on listings provider (Ars Technica)

One Last Bite 😋

Infographic: How Much Damage Do Wildfires Cause? | Statista